# Real world efficiency of a LM3150 Webench DC power supply design?

When designing a DC/DC step down converter with TI Webench with these parameters:

• Vin min: 11.4 V
• Vin max: 12.6 V
• Output : 5.0 V
• Iout : 8.0 A
• Ambient temp: 30 ºC

Then turn the efficiency knob to the right two times (this is highest efficiency) and then selecting the LM3150 buck topology design.

The resulting theoretical efficiency for 12V DC input (Vin = 12.0) is:

• 96.774% at 0.8 A load
• 98.115% at 1.6 A load
• 98.524% at 2.4 A load
• 98.696% at 3.2 A load
• 98.772% at 4.0 A load
• 98.800% at 4.8 A load
• 98.799% at 5.6 A load
• 98.781% at 7.1 A load
• 98.712% at 7.9 A load

The paper Towards a 99% Efficient Three-Phase Buck-Type PFC mentions an "error of 0.16% in efficiency" at full load between calculated and the results of calorimetric measurements.

When using exactly the components as specified in the BOM, including the prototype PCB (PC Board, Part Number 551600142-002), what will be the real world efficiency at these loads at Vin = 12.0V (instead of this theoretical simulation)?

• For practical efficiency, better hook-up the components and check.:)
– AKR
Sep 5, 2014 at 3:42

Using a National Semiconductor LM3150 evaluation board the real world measured apparent power (S in VA) efficiency is:

# Sη = 97.71% @8.00A load

The TI evaluation PCB might be designed as generic board for different voltages and not be as efficient as possible. That is why an optimised 5x5cm LM3150 PCB was designed for only a 12VDC to 5VDC conversion. The active power (P in Watts) efficiency measured using different measuring instruments, see measurement setup #2.

# Pη = 97.65% @8.00A load

Using an optimized PCB layout, efficiency results improved:

# Pη = 98.25% @8.00A load

More load versus active power efficiency results [PCB rev.1; rev.2]:

• 0.800A = 95,73%; 96,00%
• 1.000A = 96,12%; 96,38%
• 1.600A = 96,13%; n/a
• 3.200A = 97,51%; n/a
• 4.800A = 98,17%; 98,31%
• 8.000A = 97,65%; 98,25%
• 10.00A = 97,21%; 98,08%
• 12.00A = n/a; 97,89%
• 13.00A = n/a; 97,80%

### Measurement setup #1

• Vout: 5.04V
• Pout: 40.32W
• Iin: 3.50A
• Vin: 11.79V
• Pin: 41.265W
Device specifications
• Electronic load: Maynuo M9812 constant current mode; resolution 1 mA; accuracy 0.03% + 0.05% FS
• Power supply: Troniq PSU303DX2
• Multimeter (M1 current): Fluke 87 resolution 1mA; accuracy ±(0.2% + 2)
• Multimeter (M2 voltage): Fluke 17B resolution 0.01 V; accuracy ±(0.5% + 3)
Power after measurement faults
• Pin max: 41.55W
• Pin min: 40.98W
• Pout max: 40.53W
• Pout min: 40.11W
Efficiency after measurement faults
• Minimum efficiency: 96.52%
• Maximum efficiency: 98.92%

### Measurement setup #2

Device specifications for alternative PCB measurement:
• Electronic load: Itech 8512+ a.k.a. IT8512A+ constant current mode
• Power supply: Owon ODP 3032 parallel setup; constant voltage mode
• Power meter: Zes Zimmer LMG95 measure cycle: 50ms; filter: off; S-Cpl: AC+DC; internal shunt; accuracy 0.03% + 0.03% of measuring range
• Temperature: 21~25ºC
• Cable length from PCB to load and PCB to power supply: 0,4 meter each
• Cable type: stranded copper wire, diameter 1,66 mm

To see whether components matter we also compared a Webench LM2743 electronics reference design for a 12 to 5VDC power supply (including an inefficient LM2937 linear voltage regulator). The LM2743 was build in two versions of ceramic capacitors for CIN, COUTREG:

1. Murata GRM32ER61E226KE15L each € 1.51 => 90.90% efficient
2. TDK C3225X5R1E226K250AC each € 0.83 => 91.89% efficient

### Conclusion: capacitors from different brands behave different.

Component specification for the LM3150 PCB:

1. Cff = GRM2165C1H202JA01D
2. Css = GRM21BR72A153KA01L
3. M1, M2 = CSD16321Q5
4. Rilim = CRCW0805332RFKEA
5. Ron = ERJ-6ENF4643V
6. Cvcc = EMK212B7225KG-T
7. Cbst = EMK212B7474KD-T
8. Rfb1 = ERJ-6ENF1002V
9. Cbyp = 08053C104KAT2A
10. Rrb2 = ERJ-6ENF7322V
11. Cin = GRM32ER61E226KE15L
12. Cout = 16SVP330M
13. U1 = LM3150MHX/NOPB
14. L1 = SER2915L-103KL
15. PCB = 551600142-002/NOPB